Re: [Gomp-discuss] Frontend ..

[Jacob Weismann Poulsen]

Hi,

It is true that I stumbled over GOMP during the spring and that I 
spent some time browsing and hacking bits of the gcc source (first
encounter - so it is probably pretty useless) and the source available 
on the GOMP web pages. The gcc-patch is a rather straightforward dealing 
with the incomplete proposal for the gcc front-end for GOMP version 0.1, 
cf. the TODO proposal here: 
http://lists.gnu.org/archive/html/gomp-discuss/2004-04/msg00002.html

The line-based '#pragma omp' syntax check is built around the 
pragma-handler and based on the c-openmp.c proposal by Steven Bosscher, 
cf.  http://mail.gnu.org/archive/html/gomp-discuss/2003-02/msg00076.html.
I totally agree that putting the openMP pragma handling in the pragma.c
stub is probably not the right thing to do (it is quite clear from the
mail archive that Steven and others came to the same conclusion).

Nevertheless, I have learned from playing with it and others might do
the same. Moreover, the straightforward approach will give 
us room for experiments. Say that we wish to play with the data-structures 
that will be passed from the front-end on to the middle-end/back-end, e.g. 
simply try to pass some information all the way, then we have a place to 
hook these things in. That is, the state of my gcc patch is merely for 
experiments and NOT a proposal for a final front-end solution. The current 
status is that the samples from app. A (in the openMP specification)
are (hopefully correctly :)) validated on a line-based basis whenever one 
passed the 'fgomp' option to the
patched gcc.

As to the samples app. A, I had to fix most of the samples to make them 
compile, 
link and run which made me hack a bit in the auto* configuration.

After having spent some hours playing with it I mailed Biagio and asked
him how to commit patches. He gave me cvs access but I hesitated committing
it because I was unsure about how you preferred the CVS organized. I recall
mailing the list with a proposal but I think that it was never resolved.
Eventually, I mailed some of it to Scott and Biagio (I think as an
answer to reported build problems with the auto* setup).

I have by no means lost interest in the GOMP project but I went on vacation
just after I discovered (and played with) the project and at the time we 
came back we had to find a new place to live. Finding a new place to live has 
STOLEN all my spare time ever since - sorry! I plan to spend time on GOMP 
again soon and the first thing I wish to do is to give Scott some deserved 
comments on his fine proposals.

Cheers, Jacob

PS! I have attached the patch for gcc-3.3. The modifications to the gomp
project files involves renaming and moving around files. Mailing tarballs
to the list does not appeal to me. If anyone is interested I can send
them directly or we can find a place to put them.



* Biagio Lucini <address@hidden> [2004-09-07 11:00]:
> Hi Lars - 
> 
> On Monday 06 September 2004 16.16, Lars Segerlund wrote:
> >
> >   From the discusions lately I have understood that basicly we would like
> > to bring the OMP stuff into tree-ssa whithout much modification, simply
> > doing a syntax and context check but not generating something interesting.
> >
> 
> I would suggest you to start from here. First, let us make sure we can parse 
> things, then we will decide what to do with that (although I do agree that 
> all the useful stuff will go to tree-ssa). Jacob has written something which 
> was very close to a parser. Maybe we can start from there? I would like also 
> to have Jacob's opinion, since so far he has been the developer on this list 
> who got closer to have a working parser (for C).
> 
> If needed (i.e. some people are working on it) Diego could probably resync 
> the 
> branch.
> 
> Cheers
> Biagio
> 
> -- 
> =========================================================
> 
> Biagio Lucini                               
> Institut Fuer Theoretische Physik
> ETH Hoenggerberg      
> CH-8093 Zuerich - Switzerland           
> Tel. +41 (0)1 6332562  
>  
> =========================================================
> 
> 
> _______________________________________________
> Gomp-discuss mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/gomp-discuss

-- 
<address@hidden>
Fingerprint: 9315 DC43 D2E4 4F70 3AA8  F8F0 9DA0 B765 F5C8 7D26

gomp.patch
Description: Text document